Science of the Total Environment 943 (2024) 173867

Contents lists available at ScienceDirect

Science of the Total Environment

journal homepage: www.elsevier.com/locate/scitotenv

Smartphone access, digital economy, and pesticide use intensity: Evidence

from China
Lin Xie a, Zeyuan Qiu b, *, Shuyin Chen a, Xiao Lei c
a
College of Economics and Management, South China Agricultural University, Guangzhou, Guangdong, China
b
College of Science and Liberal Arts, New Jersey Institute of Technology, Newark, NJ, USA
c
Department of Sociology, Zhejiang University, Hangzhou, Zhejiang, China

H I G H L I G H T S G R A P H I C A L A B S T R A C T

• Smartphone access has a “double-edged

sword” effect on pesticide use intensity.
• Greater access led to a higher pesticide
use intensity in regions with low level of
digital economy.
• Greater access led to a lower pesticide
use intensity in regions with high level
of digital economy.
• A comprehensive social and engineering
integration is needed to reduce pesticide
use intensity.

A R T I C L E I N F O A B S T R A C T

Editor: Jacopo Bacenetti Pesticide overuse has been an increasing concern in China. Digital technology, such as smartphone access, is
considered an effective way to promote proper use of pesticides. Using the Chinese Extended Family Database
Keywords: (2015, 2017, and 2019), this study empirically examines the impact of smartphone access on pesticide use in­
Pesticide use intensity tensity among Chinese farmers. The results show a “double-edged sword” effect of smartphone access on
Smartphone access
pesticide use intensity. In rural areas with a low level of digital economy, greater smartphone access led to higher
Digital economy development
pesticide use intensity. In rural areas with a high digital economy level, smartphone access reduced pesticide use
Chinese farmers
intensity. The study results show that reducing pesticide use intensity through digital technology is not a linear
process but a complicated one that involves social and engineering integration, including an increase in access to
smartphones, development of a regional digital economy, reconstruction of agricultural extension systems, and
enhancement of the capacity of digital technology.

1. Introduction countries that attempt to modernize their agricultural economy

(Enserink et al., 2013; Popp et al., 2013; Verger and Boobis, 2013; Wu
Proper pesticide use is a challenging issue for many developing et al., 2018; Xie et al., 2020; Zhang et al., 2015a). Pesticide overuse is

* Corresponding author.
E-mail addresses: [email protected] (L. Xie), [email protected] (Z. Qiu), [email protected] (S. Chen), [email protected] (X. Lei).

https://doi.org/10.1016/j.scitotenv.2024.173867
Received 23 January 2024; Received in revised form 4 June 2024; Accepted 7 June 2024
Available online 9 June 2024
0048-9697/© 2024 The Authors. Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-
nc-nd/4.0/).
L. Xie et al. Science of the Total Environment 943 (2024) 173867

widely observed in those countries, such as China (Zhang et al., 2015a, agricultural knowledge and information easily. Video information
2015b), Thailand (Grovermann et al., 2013), Kuwait (Jallow et al., through smartphones helps farmers understand the advantages and
2017), and Vietnam (Normile, 2013; Salazar and Rand, 2020). Pesticide disadvantages of new farm technologies, ensuring accurate application
overuse leads to environmental pollution (Hao and Yang, 2013), causes (Zheng and Ma, 2023). Empirical research shows that easily accessible
food safety concerns (Zhang et al., 2021), and increases human health videos through smartphones can deliver complex and rich knowledge
risks (Enserink et al., 2013; Zhang et al., 2018; Möhring et al., 2020). and information on pesticides, increase farmers' understanding of pes­
Therefore, many technical supports and regulatory strategies have been ticides, and help farmers make more informed decisions on pesticide use
introduced to reduce pesticide use intensity and promote the proper use (Chowdhury et al., 2015). Many social media platforms allow some
of pesticides (Enserink et al., 2013; Normile, 2013; Möhring et al., experienced farmers to be video content creators and become locally
2020). influential experts (Šūmane et al., 2018). Learning through videos al­
Digital technologies have been applied to transform agricultural lows farmers to integrate such information with their local experiences
systems and offer an opportunity for properly using pesticides in many to update their knowledge and innovate their farming practices (Zossou
developing countries. Digital agriculture integrating digital technology et al., 2009).
with agricultural production helps improve agricultural resource use Although mobile phone technology, especially access to smart­
efficiency and sustainability (Rotz et al., 2019; Clapp and Ruder, 2020). phones, helps farmers increase their knowledge of pesticides and thus
Digital technologies are integrated with automation technologies such potentially reduce pesticide use, farmers in many developing countries
as intelligent tractors, drones, and robots to implement variable rate still tend to apply more pesticides due to agricultural productivity
applications that precisely deliver seeds, water, fertilizers, and pesti­ concerns (Grovermann et al., 2013; Jallow et al., 2017; Salazar and
cides and, therefore, significantly reduce the environmental impacts of Rand, 2020; Schreinemachers et al., 2020; Xie et al., 2020; Zhang et al.,
agricultural production (Clapp and Ruder, 2020). Digital technologies 2015b, 2015a). Pesticide overuse was especially apparent in China,
also reduce the cost of tracking product origins, increase transparency, where the public agricultural extension system collapsed in 1980s
improve public trust on the value chain of agricultural products, and (Zhang et al., 2015a). Without public agricultural extension services,
help promote environment-friendly production behaviors (Fabregas pesticide sellers have become the primary source of information and
et al., 2019). Through mining big data on the Internet, digital technol­ knowledge on pesticides to farmers in China (Fan et al., 2015; Jin et al.,
ogies improve farmers' communication, learning, and information 2015). The profit motive often prompts sellers to oversell agricultural
sharing (Nyamba and Mlozi, 2012; Baumüller, 2018; Zhong et al., supplies through various tactics. Since farmers in China are generally
2023). In China, businesses, such as pesticide producers Bayer and less educated and poorly trained, frequent updates on pesticide varieties
Sinochem and drone manufacturers XAG technology and DJ- and even changes in packaging further exacerbate such information
Innovations, are actively building online platforms to support farmers asymmetry between farmers and pesticide sellers (Jin et al., 2015).
adopting new pesticide use technologies. Motivated by profits, pesticide sellers often promote pesticide sales by
As a digital technology, mobile phones play a key role in the digital overemphasizing the role of pesticides in agricultural production,
transformation of agriculture (Min et al., 2020). Mobile phone tech­ resulting in excessive purchase and overuse of pesticides in China (Wang
nology has become increasingly adopted to promote agricultural and Gu, 2013). Although mobile phones help farmers obtain market
extension services in developing countries (Aker, 2011; Birke et al., information about pesticides (Cole and Fernando, 2021; Freeman and
2019; Cole and Fernando, 2021; Folitse et al., 2019; Janc et al., 2019). Qin, 2020; Tambo et al., 2023), the same technology can also be used by
Mobile phones help disseminate product information and increase the pesticide sellers to directly market their products to farmers, which leads
adoption of the recommended agrochemical application rate by 22 % in to pesticides overuse. A survey from farmers in China suggested that
sub-Saharan Africa and India (Fabregas et al., 2019). Mobile phones also mobile phone technology, especially smartphone access, has a double-
connected the ICT revolution to the smallholder cotton farmers and edged sword effect on pesticide use: smartphone access resulted in a
helped increased the adoption of pesticide use recommendations and 33 % increase in pesticide expenditure among the farmers in the lowest
biological pest control methods through an innovative voice-based ICT 20th quantile of the pesticide expenditure, and a 36–39 % reduction in
advisory service in India (Cole and Fernando, 2021). In Uganda, where pesticide expenditure among the farmers in the upper 60th and 80th
resources are scarce, mobile phones help farmers to access information quantiles of pesticide expenditure (Ma and Zheng, 2022).
on agricultural inputs including pesticides and increase the likelihood of However, no study has yet attempted to systematically analyze the
using these inputs (Freeman and Qin, 2020). In Rwanda and Uganda, inherent cause of such a double-edged sword effect of mobile phone
mass social media campaigns involving the use of mobile phones technology on pesticide use. This study aims to understand the under­
effectively improved farmers' awareness of pesticide risks and safety lying impact of smartphone access on pesticide use in China. The hy­
measures and boosted the adoption of environmentally safer alterna­ pothesis is that the development of a regional digital economy where
tives to synthetic pesticides (Tambo et al., 2023). In Ecuador, mobile mobile phones operate significantly impacts how smartphone access
phone applications helped increase the adoption of low-toxicity pesti­ affects pesticide use. Digital learning through smartphones relies on the
cide products among the blackberry farmers (Carrión-Yaguana et al., smooth operation of a local digital network characterized by an excel­
2020). lent digital infrastructure, a large population of internet users, and
The impact of mobile phones on farmers' behaviors in agricultural active content creation. For example, the expansion of regional mobile
input use is not uniform and has been highly affected by the develop­ phone coverage induces farmers' market participation (Muto and
ment of the technology itself. Early mobile technology relied on text Yamano, 2009) and increases the provision of agricultural extension
messages in the 2G era and pictures, graphics, voice messages in the 3G services (Aker, 2011). Without the support of a well-developed local
era. These traditional mobile technologies were not able to provide so­ digital network, smartphones are no better than traditional feature
phisticated advice on agricultural practices or new technologies in many phones, and therefore, farmers with smartphones cannot reap the ben­
developing countries (Awuor and Otanga, 2019; Nakasone et al., 2014). efits of digital learning. For example, in extremely remote rural com­
The primary reason is that reading and interacting with text messages munities in Papua, Indonesia, smartphones were mainly used for calls
through mobile phones requires a high level of literacy that smallholder and text messages (Heimerl et al., 2015).
farmers do not have (Nakasone et al., 2014; Wyche and Steinfield, Based on longitudinal survey data on rural households in China, this
2015), and eventually leads to the problem of information overload study assesses how smartphone access and the development of the
(Awuor and Otanga, 2019). regional digital economy jointly affect farmers' decisions on pesticide
However, smartphone technology with video message functions in use, which explains the double-edged sword effect of smartphone access
the 4G and 5G eras allows farmers with a limited literacy to obtain on pesticide use among Chinese farmers. As a developing country where

2
L. Xie et al. Science of the Total Environment 943 (2024) 173867

smartphones are widely used, China provides an interesting example to 2.2. The growth of mobile communication in China
understand the impacts of smartphones and mobile technology on
pesticide use in agriculture. Like many other developing countries in China launched its first commercial service of the first-generation
South Asia and Africa, agriculture in China is still dominated by small­ analog mobile communication system (1G) in 1987. Over the next 35
holder farmers. Therefore, the findings on the impact of smartphones on years, China has steadily upgraded its mobile communication system
pesticide use in China can help developing countries to develop effective from 1G to 5G and currently owns the largest 5G network in the world.
policies and strategies to properly use pesticides and other According to the China Ministry of Industry and Information Technol­
agrochemicals. ogy, China had 475 million 5G mobile phone users, which accounted for
>70 % of the global 5G network users by July 2022. China operates 1.97
2. Research background million 5G base stations, accounting for over 60 % of the active 5G base
stations worldwide. The growth of mobile communication in the past 35
2.1. Pesticide overuse in China years in China is further evidenced by a sharp rise in mobile phone
penetration rate, a swift conversion to smartphones, and a steady in­
In traditional agriculture in China, farmers rely on intensive culti­ crease in mobile internet speed.
vation techniques such as timely planting and harvesting, weeding, and Fig. 2a shows the penetration rates, i. e., the number of phones per
deep plowing to physically prevent and control plant diseases and pests 100 people, of hardline phones and mobile phones in China from 2000
(Wen, 2016). Despite the fast growth of the “green revolution” in to 2021. Over the 20 years, the mobile phone penetration rate had risen
Western developed countries and other parts of the world, China limited from 6.7 to 116.3, while the penetration rate for the hardline phones
manufacturing capacity and use of synthetic fertilizers and pesticides increased from 12.4 in 2000 to 28.1 in 2006 and then decreased to 12.8
until the 1980s (Pan et al., 2020). Agrochemicals, including fertilizers in 2021. By the end of 2021, there were 1.82 billion telephone users in
and pesticides, grew in popularity in the middle of the 1980s for two China, including 1.64 billion mobile phone users and only 181 million
reasons. First, China began to implement its “open door” policies to hardline phone users.
foreign technology and investment, significantly increasing the capacity Fig. 2b shows the growth of smartphones with 3G/4G services in
of manufacturing and supplying agrochemicals in China. Second, China China from 2010 to 2018 in terms of the number of phone users, the
widely implemented the Rural Household Contract System to grant ratio of the smartphone users among all mobile phone users, and the
farmers nearly complete freedom to conduct their agricultural opera­ ratio of the number of 3G/4G base stations among all mobile base sta­
tions, which gave farmers the freedom to use agrochemicals to increase tions. The number of 3G/4G users was only 47.05 million, accounting
crop yields and farm income. for 5.5 % of the total mobile phone users in 2010. It grew to 1.3 billion in
Fig. 1a presents the total amount of pesticide usage (the left axis) and 2018, i. e., 83.4 % of mobile phone users. The number of 3G/4G base
the pesticide use intensity on the sown area of crops in China from 1991 stations accounted for 32.8 % of all base stations in 2010 and had grown
to 2021. The total pesticide use in China increased from 0.77 million to 75.5 % by 2018.
tons in 1991 to 1.32 million tons in 1999, a nearly 73 % increase. The As a result, the number of mobile internet users has also grown
total pesticide usage dropped slightly in 2000, and resumed steady rapidly in China. According to the China Internet Network Information
growth in 2001, reaching its peak of 1.81 million tons in 2013. The Center, the number of mobile internet users in China grew from 50.4
intensity of pesticide use in China experienced similar changes over that million in 2007 to 1.03 billion in 2021, corresponding to 24 % of the
period and reached its peak level of 11.14 kg/ha in 2013. The fast total internet users in 2007 and 99.7 % in 2021. In other words,
growth in pesticide use has become a significant concern in China. The smartphones have become essential for people in China who want to
Chinese government began to take a series of actions to reduce pesticide access the Internet.
use in 2010. The Chinese Ministry of Agriculture issued the Action Plan The speed of downloading and uploading mobile Internet has also
on Agricultural Crop Pest Control and Disease Prevention in 2010 to increased rapidly. Fig. 2c presents the average downloading speed of the
prevent crop diseases and control crop damage while reducing pesticide 4G network from 2016 Q3 to 2021 Q4. Mobile internet speed has sky­
use and enhancing agricultural product safety. The Central Government rocketed since the launch of the 5G network in China. According to the
in China subsequently issued its pesticide reduction policy in the 14th China Information and Communication Academy, the average down­
Five-Year Plan for National Economic and Social Development of the loading speed for 5G users in China was 341.2 Mbps, and the average
People's Republic of China and the Outlook of the 2035 Long-term uploading speed was 71.98 Mbps in 2022, among the world's fastest
Development Goals. These policy changes on pesticide use and their mobile networks. Mobile smartphones are predominantly used for
implementation had impacts. As shown in Fig. 1a, both the total pesti­ instant messaging, video sharing, and watching short movies.
cide usage and the intensity of pesticide use declined steadily after 2013, Despite the staggering growth of mobile communication in China,
and lowered to 1.24 million tons and 7.35 kg/ha, respectively, in 2021. China's rural regions are lagging. According to the China Internet
Fig. 1b compares the intensity of pesticide use measured by the ratio Network Information Center, the internet penetration rate was 58.8 % in
of the average weight of active ingredients per unit of farmland in kg/ha the rural regions and 82.9 % in urban and urbanizing regions in China in
in China to the world from 1990 to 2021, based on data from FAO.1 2022. The internet speed in rural regions is far slower than in urban and
From 1990 to 1992, the intensity of pesticide use in China was lower urbanized regions.
than the world average, as the ratio was less than one. However, from
1993 to 2018, the intensity of pesticide use in China increased signifi­ 3. Method
cantly and surpassed the world average level. The peak was in 2009
when the intensity of pesticide use in China was approximately 1.36 3.1. Data
times the world average. The intensity of pesticide use in China started
to drop below the world average in 2019. This research uses the longitudinal survey data from the Chinese
Family Database (CFD) at Zhejiang University and the China Household
Finance Survey (CHFS) conducted by the Survey and Research Center
for China Household Finance at China Southwestern University of
Finance and Economics (SWUFE). The CFD survey adopted a three-
stratified probability proportion to size (PPS) sampling approach to
1
Source: FAO. Pesticides Use. FAOSTAT, URL. https://www.fao. select its household samples for the longitudinal survey (Wu et al.,
org/faostat/zh/#data/RP 2018). The selected samples include households from 29 provinces,

3
L. Xie et al. Science of the Total Environment 943 (2024) 173867

Fig. 1. Pesticide use intensity comparisons in China and the World.

Fig. 2. The mobile communication infrastructure growth patterns in China.

autonomous regions, and metropolitan areas directly under the control survey data to form a non-equilibrium panel data. The CFD survey
of the central government in China. Through scientific sampling, mod­ contains the essential information on farmers' farming operations,
ern survey techniques, and rigorous survey management, the longitu­ including pesticide use and the information on mobile phones, the
dinal surveys aim to collect high-quality micro information on Chinese Internet, and other digital technologies, and is used in this study.
families to conduct in-depth research on various issues related to Chi­
nese families. Among them, the 2015 data involved a rural sample of
22,535 households; The 2017 data involved a rural sample of 24,764 3.2. Variables
households; The 2019 data involved a rural sample of 21,815 house­
holds. This study extracted all rural household samples growing staple 3.2.1. Dependent variable
food and/or cash crops from the CFD database, which included 5054 in This study uses pesticide use intensity as the dependent variable. The
2015, 9142 in 2017, and 8757 in 2019. We use these three rounds of quantity and expenditure of pesticides applied per unit area can be used
to define pesticide use intensity, but the CFD survey only reported the

4
L. Xie et al. Science of the Total Environment 943 (2024) 173867

quantity of pesticide applied in 2015, and not in other years. Therefore, Table 1
we define the pesticide use intensity as the ratio of the total pesticide Description of variable indicators.
expenditure to the sown area following other similar studies (Zhu and Variable category and name Questions in the Variable References
Wang, 2021). This study takes the logarithmic value of the average CFD Surveys specification
pesticide expenditure to reduce the impacts of those extreme values on Dependent Pesticide use How much did Ln (pesticide (Ma and
the results. variable intensity your family expenditure / Zheng,
spend on the sown 2022; Wu
3.2.2. Variables of interest pesticides area + 1) et al., 2018)
(yuan)? What is
This study includes two independent variables as the variables of your family's
interest. The first is access to the smartphone. Farmers in China mainly sown area for
access digital technology services through mobile Internet and smart­ staple food and
phones. The CFD contains specific questions on the use of mobile phones cash crops?
Independent Smartphone What kind of One if “1” is (Khan et al.,
among farmers, including the specific types of mobile phones being
variable mobile phone selected, and 2022; Ma
used. Following the similar studies (Khan et al., 2022; Ma and Zheng, are you using zero and Zheng,
2022), this study uses a dummy variable to indicate farmers' smartphone now? (1. otherwise 2022)
access: one if using a smartphone and zero otherwise. The second is the Smartphone; 2.
digital economy index (DEI), which measures regional differences in the Non-
smartphone; 3.
development in digital economy in China. DEI is derived from five in­ No cell phone)
dicators: internet penetration rate, internet-related employment, Digital As described in In the range (Zhao et al.,
internet-related economic output, mobile internet users, and digital in­ economy the text between 2020)
clusive finance index. China National Bureau of Statistics compiles the index 0 and 1
Control Age The year of The year in (L. Fan
first four indicators, and the last one is jointly compiled by the Digital
variable birth of the which the et al., 2015;
Finance Research Center of Peking University and the Ant Financial farmers survey was Pan et al.,
Services Group. The five indicators were further standardized and interviewed conducted 2021;
aggregated to calculate the composite DEI of all 30 provinces, autono­ minus the Zheng
mous regions, and metropolitans under the control of the Central Gov­ year of birth et al., 2021;
of the farmer Zhou and
ernment following Zhao et al. (2020).
interviewed Jin, 2009)
Education What is the The actual
3.2.3. Control variables literacy level of years of
The control variables include the characteristics of rural households, the farmers education for
interviewed?? options 1–9
villages, and farmland that potentially affect farmers' decisions on
1. Never went are assigned
pesticide use. The control variables on the rural household include to school; 2. 0, 6, 9, 12, 9,
farmers' age (Al Zadjali et al., 2014; Huang et al., 2020; Zhou and Jin, Primary school; 12, 16, 19,
2009), gender (Atreya, 2007; Wang et al., 2017), education (Al Zadjali 3. Junior high and 23
et al., 2014; Goodhue et al., 2010; Jallow et al., 2017; Zhou and Jin, school; 4. High
school; 5.
2009), agricultural employment ratio (Hedlund et al., 2020), whether
Technical
the household has any agricultural loan (Rahman, 2003), and received secondary
any agricultural guidance (Huang et al., 2021), and production for self- school; 6.
consumption (Sharifzadeh et al., 2019). The control variables for the Junior college;
7. Bachelors; 8.
village include per-capita income (Li et al., 2023)2 and the distance of
Masters; 9. Ph.
the village to the nearest market (Huang et al., 2020). The control var­ D. s
iables on farmlands include the total sown area (Wu et al., 2018), Gender The gender of Select one for
planting structure (the ratio of the sown area for food crops) (Rahman, the farmers female, zero
2016), and land ownership (farmland certification) (Migheli, 2017). The interviewed otherwise
Ratio of How many The number
study also contains the control variables on the household and year fixed
agricultural members of of farm
effect being surveyed using dummy variables. employment your family are laborers/ the
Table 1 presents detailed information on those variables, including engaged in workforce
the variable type, variable name, relevant survey questions, and calcu­ agricultural size
production and
lating specifications and relevant references for using the variables.
operation?
Ratio of What are the The sown
3.3. Fixed effect models staple food sown areas for area for staple
crops staple food food crops/
crops? the total sown
Following Greene (2003) and Wooldridge (2010), this study applies area of crops
a two-way fixed effect model to assess the effect of smartphone access on Agricultural Does your One if “1” is
pesticide use intensity in China. The two-way fixed effect model was loan family have selected, zero
outstanding otherwise
used in other studies using the same database being used in this study
bank loans for
(Fan et al., 2023; Zheng et al., 2023a, 2023b; Zhu and Wang, 2021). agricultural
Model 1, specified as Eq. (1), considers only the impact of smartphone operations? (1.
access on pesticide use intensity: Yes; 2. No)
Farmland Does your One if “1” is
Certification farmland have a selected, zero
rural land otherwise
2
The existing literature mainly discussed the relationship between household contract right
income level and pesticide application. However, given the endogeneity of certificate?
household income, we controlled the village's annual per capita income as an (continued on next page)
alternative. This variable is a higher-level variable that helps reduce the
endogeneity.

5
L. Xie et al. Science of the Total Environment 943 (2024) 173867

Table 1 (continued ) where x2it is the provincial DEI where farmer i is located in year t, β2 is
Variable category and name Questions in the Variable References the regression coefficient for independent variable x2it and indicates if
CFD Surveys specification the regional development of the digital economy impacts the pesticide
Production What were the One if “3”, use intensity. Other variables are previously defined.
for self- main uses of the “4” is Model 3 considers the compounding impacts of both independent
consumption agricultural selected, zero
variables by adding an interaction of the two independent variables to
products your otherwise
family Model (2) and is specified as follows in Eq. (3):
produced last
yit = α + β1 x1it + β2 x2it + β3 x1it x2it + γZit + μi + νt + εit (3)
year? 1. Direct
sales; 2.
Processed and where β3 is the regression coefficient for the newly created interaction
sold; 3. For your term x1it x2it and indicates the compounding impacts of smartphone
use; 4. For your access and the regional development of the digital economy on the
agricultural
pesticide use intensity, and other variables are previously defined.
production; 5.
Not yet
produced; 6. To
be sold; 7. 3.4. Endogeneity test
Other.
Agricultural Has your family One if “1” is
guidance received any selected, zero The fixed effects model is subject to some endogeneity concerns
technical otherwise because of the potential reverse causal relationship between smartphone
guidance on access and pesticide use intensity and potential omission of other critical
agricultural control factors in our model specification. We tried to address potential
production? (1.
endogeneity concerns by incorporating an instrumental variable in our
Yes; 2. No)
Total sown What is the Ln (The total models. Following Evans et al. (1992), we introduced an instrumental
area sown area for sown area + variables at a higher aggregate level to our models to address the
both staple food 1) endogeneity concerns. We select the smartphone usage rate in the
crops and cash
village level as the instrumental variable for smartphone usage. This
crops?
Annual per What is the per ln (Per capita
usage rate is the result of excluding the sample itself. The smartphone
capita capita income of the rate in the area reflects the prevalence of digital technology applica­
income of the disposable village +1) tions. Under the influence of neighborhood effects, individuals in the
village annual income areas with higher smartphone rates are more likely to access smart­
of the residents
phones, thereby meeting the requirement of instrumental variable
in this
community? relevance. However, the smartphone usage rate in a village does not
Distance What is the Ln (Distance affect individual decision on pesticide use, satisfying the requirement of
from market distance from from market instrumental variable exogeneity.
the village to +1), if the
the nearest farmers'
farmers' market market or free 4. Results
or free market? market is in
the village, 4.1. Descriptive statistics of all variables
the distance is
0
Household Household who Dummy – In this study, we restricted the age of the surveyed farmers to 16 to
was variable 80 years, as farmers who were too young or too old could not make their
interviewed own decisions. To reduce measurement error, following Wu et al.
Year The year when Dummy
(2018), we excluded households spending less than one yuan (RMB) on
the survey was variable
conducted pesticide inputs per mu of sown area in any given year and/or did not
report any sown area. Table 2 shows the descriptive statistics of all
variables used in the study based on the combined rural household
yit = α + β1 x1it + γZit + μi + νt + εit (1) samples in 2015, 2017, and 2019. The household expenditure on
pesticide use per hectare of sow area in a year ranges from ￥15 to
where i is the index for the farmers being surveyed, t indicates the year ￥28,350. On average, a rural household uses ￥1440.17 per hectare of
when the survey was conducted, yit is the pesticide use intensity for the sown area. The average smartphone value was 0.4071, indicating
farmer i in year t, x1it indicates whether farmer i uses a smartphone in that 40.71 % of rural households use smartphones. The average digital
year t, Zit is a vector of all control variables for farmer i in year t, α is the Economy index is 0.3421 on a scale of 0 to 1.
intercept, β1 is the regression coefficient for the independent variable The use of mobile phones, especially smartphones, among the
x1it, and γ is a vector of coefficients for the control variable vector Zit , selected rural households steadily increased over the sampling period.
and μi is the household fixed effect, and νt is the time-fixed effect and ϵit Among all selected rural households in this study, only a small per­
is the residual term. The coefficient β1 Eq. (1) indicates whether farmers' centage didn't have mobile phones, i.e., 3.8 % in 2015, 3.4 % in 2017,
smartphone access impacts pesticide use intensity. and 2.4 % in 2019, and the number gradually declined over the sampling
As discussed previously, smartphones rely on the availability and periods. Among the selected rural households, 70 % used non-
efficiency of the digital network to have the expected impacts on smartphones, and only 26.2 % used smartphones in 2015. The con­
farmers' behavior change on pesticide application. We also introduce the struction of 4G network base stations and the introduction and growth of
DEI as an independent variable in the analysis to assess such effects. 4G smartphones accelerate the adoption among rural households after
Model 2 considers the impacts of DEI by directly adding the DEI as an 2015. The percentage of selected rural households using smartphones
independent variable as specified in Eq. (2): increased to 44 % in 2017 and 57.6 % in 2019, while the percentage
yit = α + β1 x1it + β2 x2it + γZit + μi + νt + εit (2) using non-smartphones was decreased to 52.5 % in 2017 and 40 % in
2019.

6
L. Xie et al. Science of the Total Environment 943 (2024) 173867

Table 2
Statistical description and analysis results of each variable.
Variables Sample size Mean SD. Min Max
− 1
Pesticide use intensity (Yuan⋅ha ) 10,255 1440.17 2667.85 15 28,350
Pesticide use intensity (Yuan⋅ha− 1, logarithm) 10,255 6.4600 1.2291 2.7726 10.2524
Smartphone 10,343 0.4071 0.4913 0 1
Digital economy index 10,343 0.3421 0.0834 0.2332 0.7727
Age 10,343 54.3781 11.3548 16 80
Education 10,338 8.3405 4.5862 0 19
Gender 10,343 0.6365 0.4810 0 1
Ratio of agricultural employment 10,343 0.6133 0.3428 0 1
Ratio of staple food crops 10,343 0.9253 0.1865 0 1
Agricultural loan 10,327 0.0620 0.2411 0 1
Farmland certification 10,343 0.6462 0.4782 0 1
Production for self-consumption 10,301 0.4615 0.4985 0 1
Agricultural guidance 10,337 0.0928 0.2901 0 1
Total sown area (ha) 10,274 0.7363 1.0735 0.0007 10.6667
Total sown area (ha, logarithm) 10,274 0.4532 0.3892 0.0007 2.4567
Annual per capita income of the village (Yuan) 10,160 9,815.179 8,751.374 0 60,000
Annual per capita income of the village (Yuan, logarithm) 10,160 8.8151 0.9876 0 11.0021
Distance from market (km) 9,065 4.9320 6.0219 0 60
Distance from market (km, logarithm) 9,065 1.4300 0.8289 0 4.1109
IV 10,213 0.4062 0.2520 0 1

Note: The raw value minus the sample mean was used for regression analyses to avoid multicollinearity problems following Balli and Sørensen (2013). The sample size
is obtained after excluding observations with missing values.

4.2. Fixed effect modeling results

Table 3
The fixed effect modeling results excluding the top 1 % of households in pesti­
We further excluded the top 0.1 %, 0.5 %, 1 % or 5 % of households
cide use intensity, total sown area, and agricultural income per capita.
in pesticide use intensity and total sown area to eliminate the potential
Variables Model 1 Model 2 Model 3
impacts of sampling outliers to run the fixed effect models. The results
are consistent. Table 3 presents the modeling results for the fixed effect Smartphone 0.0770* 0.0787* − 0.0085
models excluding the top 1 % of households in pesticide use intensity (0.0412) (0.0412) (0.0518)
Digital economy index (DEI) − 3.4017** − 1.2794
and total sown area while the supplemental materials presents the re­
(1.6443) (1.8442)
sults excluding the top 0.1 %, 0.5 % and 5 % of households in pesticide Smartphone * DEI − 3.1269***
use intensity and total sown area. Based on Model 1, the coefficient for (1.0853)
the independent variable smartphones is 0.077 and statistically signifi­ Age 0.0000 − 0.0002 − 0.0003
(0.0030) (0.0030) (0.0030)
cant at a 90 % confidence level. This result indicates that smartphone
Education − 0.0006 − 0.0006 − 0.0004
access would increase pesticide use intensity. Following the calculation (0.0046) (0.0046) (0.0046)
methods in Kennedy (1981), the estimated coefficient of 0.077 indicates Gender 0.0184 0.0188 0.0141
that the average expenditure on pesticide use per ha is 7.9 %3 higher for (0.0519) (0.0519) (0.0519)
rural households using smartphones than those not using smartphones, Ratio of agricultural employment 0.0494 0.0581 0.0577
(0.0571) (0.0571) (0.0568)
including non-smartphones and not having phones. The modeling re­
Ratio of staple food crops 0.4386*** 0.4328*** 0.4379***
sults from Model 2 confirm the findings from Model 1 on the significant (0.1120) (0.1114) (0.1111)
and positive impacts of smartphone access on pesticide use intensity. Agricultural loan 0.0731 0.0696 0.0661
The results from Model 2 also indicate that the reverse relationship (0.0795) (0.0795) (0.0795)
Production for self-consumption − 0.0646 − 0.0673* − 0.0711*
between the digital economy index and the pesticide use intensity as the
(0.0400) (0.0399) (0.0400)
model coefficient for DEI is negative at a 95 % confidence level. Farmland certification − 0.0507 − 0.0532 − 0.0572
As shown in the modeling results from Model 3, the coefficients for (0.0383) (0.0384) (0.0381)
smartphone and DEI have the same signs as those from Models 1 and 2, Agricultural guidance 0.0269 0.0271 0.0235
but they are no longer statistically significant. However, the coefficient (0.0554) (0.0554) (0.0551)
Total sown area − 1.5150*** − 1.5125*** − 1.5118***
for the cross-multiplication term between smartphone and DEI, which
(0.1001) (0.1000) (0.1000)
indicates the compounding impacts of smartphone access and the − 0.0277 − 0.0278 − 0.0285
development of the digital economy, is negative and statistically sig­ (0.0187) (0.0187) (0.0186)
nificant at a 99 % confidence level. In other words, smartphone access Distance from market − 0.0274 − 0.0252 − 0.0266
with a growing digital economy could lead to a decline in the intensity of (0.0195) (0.0197) (0.0196)
Constant 7.1688*** 7.1011*** 7.1705***
pesticide use. (0.2711) (0.2726) (0.2741)
Table 3 also shows the consistent impacts of all control variables Household fixed effect Yes Yes Yes
across the three models. A higher ratio of agricultural employment and a Year fixed effect Yes Yes Yes
larger scale of sown areas dedicated to staple food crops lead to higher N 7998 7998 7998
Adjusted R2 0.1787 0.1800 0.1824
pesticide use intensity, as indicated by their positive and statistically
F 25.5976 24.3475 23.2594

Note: Significance levels: * p < 0.1, ** p < 0.05, and *** p < 0.01. The sample
size is obtained after excluding observations with missing values.
( )
3
The calculation formula is: g * = exp ̂c − 1̂
2 V (̂
c) − 1.where g* is the rela­
significant coefficients. The statistically positive coefficients indicate
tive effect on pesticide use intensity of the presence of the factor represented by that carrying agricultural loans prompts farmers to increase pesticide
the smartphones, ̂c is the estimate of the regression coefficient, and V(̂
̂ c ) is the use intensity. However, farmers who consume their own products tend
variance of ̂c .

7
L. Xie et al. Science of the Total Environment 943 (2024) 173867

to lower pesticide use intensity, as shown by significantly negative co­ intensity than those without smartphone access. Our findings suggest
efficients at the 99 % confidence level. A larger scale of operation that not only smartphone access but also the development in digital
indicated by the total sown area leads to lower pesticide use intensity, as economy significantly affects the intensity of pesticide use. Smartphone
indicated by negative and statistically significant coefficients. The co­ access in the regions with a higher level of development in digital
efficients for other control variables, including farmers' age, gender, economy helps lower the pesticide use intensity in China. The overall
years of education, farmland certification, the annual per capita income increase in the intensity of pesticide use associated with smartphone
of the village, and distance from the market, don't significantly impact access in China was primarily due to the significant increases in pesti­
pesticide use intensity as their coefficients are not statistically cide use intensity in those regions with lower levels of development in
significant. digital economy. The results are robust, as the regression coefficients
Based on the fixed effect results in Model 3, we also plot the inter­ from the fixed effect models are fairly consistent.
active impacts of smartphone access and the DEI on pesticide use in­ Empirical results indicate that smartphone access must be accom­
tensity in Fig. 3, where the x-axis represents smartphone access and the panied by a high level of development in digital economy to reduce
y-axis represents the pesticide use intensity. Following Li and Sun farmers' pesticide use. However, the regions with a low development
(2017), these impacts were evaluated at the sample means of all control level of digital economy in China generally have insufficient digital
variables, and the high and low digital economy development levels infrastructure, which results in poor network communication, weak
were defined as the top and bottom 10 % DEI levels of the samples, network signal coverage, slow network speed, and nearly no existence of
respectively. As shown in Fig. 3, the pesticide use intensity for farmers content creation and other supporting services. Without the public
with smartphone access is higher than for farmers without smartphone agricultural extension services, pesticide sellers have been the primary
access under the low digital economy development scenario but higher source for most farmers to gain pesticide product information and
under the high digital economy development scenario. Also, the in­ application knowledge in China (Fan et al., 2015; Jin et al., 2015).
tensity of pesticide use under the high digital economy development Because of the lack of digital infrastructure, the farmers with smart­
scenario is lower than under the low digital economy development phone access still mainly obtain information through phone calls and
scenario. The results in Fig. 3 confirm the double-edged sword effect of SMS text messages, as in the case of Papua, Indonesia (Heimerl et al.,
smartphone access on pesticide use intensity observed by several 2015). The CFD also shows that households with smartphone access are
empirical studies in China. financially better off and socially better connected than those without
smartphones. Therefore, they are more willing and able to take risks and
increase pesticide use intensity.
4.3. Endogeneity test results However, in regions with a high level of development in digital
economy, smartphones are a new tool in agricultural operations and
Table 4 presents the endogeneity test results using the panel data. We profoundly impact agricultural productivity. Farmers can rely on
follow the mature method and perform the first stage estimation to test smartphones to independently search and access online information on
the correlation between the instrumental and endogenous variable resources and products related to agricultural production. Diverse forms
(Wooldridge et al., 2016). Based on the first stage estimation, the co­ of information such as text, pictures, voice, and video also help farmers
efficient for the instrumental variable is 0.1331 and is statistically sig­ break through many traditional barriers imposed by limited education,
nificant at a 99 % confidence level. This result indicates that the lack of social connections, and spatial isolation to learn more about
smartphone usage rate in the village would increase smartphone access crops and pesticides so that they can more properly use the products to
and shows that the instrumental variable satisfies the requirement of improve the efficiency of pesticide use and, in many cases, reduce
correlation with the endogenous variable. pesticide use. The flow of knowledge and information on social media
The second stage confirms the findings on the significant and positive then stimulates the growth of local digital services and content creation
impacts of smartphone access on pesticide use intensity. According to in the rural areas in those regions. Many experienced farmers become
previous literature (Stock and Yogo, 2002), we utilize the squared t- content creators to share their experiences and knowledge of using
statistic of the regression coefficient for the instrumental variable from pesticides through short videos, live broadcasts, online courses, and
the first-stage regression, denoted as the F-value, to determine whether other means (Šūmane et al., 2018). Such localized content and services
there is a weak instrument problem. The results indicate that the F-value have been proven to be very effective in influencing farmers' decision in
is 11.297,4 which exceeds 10, suggesting that this study has no weak pest control and disease prevention (Carolan, 2022).
instrument problem. The findings of this study have important implications for achieving
China's pesticide reduction goals. First, smartphone access has a great
5. Discussion potential to reduce the intensity of pesticide use; however, the potential
can only be realized with the development of the local digital economy
The understanding of the impacts of smartphone access on pesticide and the support of digital infrastructure. Therefore, to reduce the in­
use intensity has been paradoxical. Some research argued that the tensity of pesticide use, governmental policies should focus on
integration of smartphones into agricultural production and manage­ increasing farmers' access to smartphones and other smart terminals
ment allows farmers to access agricultural knowledge and product in­ such as tablets and investing in digital infrastructure to promote the
formation to develop smart and precision agriculture and offers development of the local digital economy. Second, after China decom­
opportunities to properly use pesticides and achieve pesticide reduction missioned its public agricultural extension system in the 1980s, agri­
goals (Rotz et al., 2019; Clapp and Ruder, 2020). However, some cultural resource suppliers essentially filled the void left by the public
empirical studies found that smartphone access did not reduce but, in extension system and became the sole source of agricultural knowledge
some cases, even increased pesticide use intensity (Ma and Zheng, and product information for farmers who seek advice, which causes
2022). pesticide overuse in China (Wang and Gu, 2013). In this digital trans­
Our empirical assessment results explain such a paradox. Overall, formation era, Chinese governments must build the public digital agri­
smartphone access hasn't helped reduce pesticide use in China yet: rural cultural extension system by investing in app development, content
households with smartphone access had 8 % higher pesticide use creation and network building to disseminate scientific knowledge on
pesticides and their proper uses. The governments may also encourage
more experienced farmers to get involved in media content creation and
4
( )2
become technical experts in local digital communities to promote peer
The calculation formula is: F = t 2 = 0.1331
= 11.297.
0.0396 learning and knowledge sharing. Third, the farmers need to be trained

8
L. Xie et al. Science of the Total Environment 943 (2024) 173867

Fig. 3. The impacts of digital economy development on pesticide use intensity.

6. Conclusion
Table 4
Endogeneity test results excluding the top 1 % of households in pesticide use
This study examines the relationship between smartphone access and
intensity, total sown area, and agricultural income per capita.
pesticide use intensity among rural households in China based on the
(1) (2)
Zhejiang University Chinese Family Database in 2015, 2017, and 2019.
Smartphone Pesticide use intensity The results show that smartphone access increased pesticide use in­
Smartphone 1.3570* tensity among the sampled rural households in China in those periods.
(0.7327) However, the impacts were mediated by the level of development in
Instrumental Variable 0.1331*** regional digital economy and vary across different regions in China. In
(0.0396)
the regions with low levels of development in digital economy, smart­
Age − 0.0182*** 0.0235*
(0.0013) (0.0137) phone access intensifies pesticide use. However, in the regions with high
Education 0.0057*** − 0.0075 levels of development in the digital economy, smartphone access helps
(0.0021) (0.0067) reduce the intensity of pesticide use. Rural households in rural regions
Gender 0.0446* − 0.0574 with a high level of development in digital economy can use smart­
(0.0228) (0.0657)
phones to access all different contents on pesticides and their proper use,
Ratio of agricultural employment 0.0326 0.0158
(0.0269) (0.0699) develop knowledge on crop disease prevention and pest control, and
Ratio of staple food crops − 0.0114 0.4232*** ultimately reduce the pesticide use intensity. The study results demon­
(0.0444) (0.1096) strate the potential of digital technologies, such as smartphone access, to
Agricultural loan − 0.0055 0.0867
reduce the intensity of pesticide use. However, substantial work is
(0.0344) (0.0847)
Production for self-consumption − 0.0153 − 0.0457 needed to realize such potential. This work includes developing regional
(0.0194) (0.0491) and local digital infrastructure, rebuilding the public agricultural
Farmland certification − 0.0131 − 0.0267 extension services, and foundational training on farmers to use
(0.0174) (0.0441) smartphones.
Agricultural guidance 0.0288 − 0.0029
(0.0265) (0.0682)
Total sown area 0.0761** − 1.6134*** Funding disclosure
(0.0349) (0.1030)
Annual per capita income of the village − 0.0052 − 0.0234 Lin Xie acknowledged a partial funding support by the National
(0.0088) (0.0219)
Social Science Fund of China (21BJY184). Zeyuan Qiu acknowledged a
Distance from market − 0.0018 − 0.0247
(0.0094) (0.0231)
partial funding support by the U.S. National Science Foundation (NSF,
Household fixed effect Yes Yes CBET-1903597). However, the funding agencies had no role in study
Year fixed effect Yes Yes design, data collection and interpretation, or the decision to submit the
N 7900 7900 work for publication.
R2 0.2843 0.0805
F 54.49 –
CRediT authorship contribution statement
Note: Significance levels: * p < 0.1, ** p < 0.05, and *** p < 0.01. The sample
size is obtained after excluding observations with missing values. Lin Xie: Writing – review & editing, Writing – original draft, Su­
pervision, Project administration, Methodology, Formal analysis, Data
on how to use smartphones to access the necessary information since the curation, Conceptualization. Zeyuan Qiu: Writing – review & editing,
general education level of the farmers is low in China. According to the Writing – original draft, Validation, Supervision, Methodology, Formal
China Internet Network Information Center, a lack of user skills, even analysis, Data curation, Conceptualization. Shuyin Chen: Methodology,
the basic keyboard input skills, often prevents many rural residents from Formal analysis, Data curation. Xiao Lei: Methodology, Formal analysis,
accessing and effectively using the Internet. Data curation.

9
L. Xie et al. Science of the Total Environment 943 (2024) 173867

Declaration of competing interest Hao, G., Yang, G., 2013. Pest control: risks of biochemical pesticides. Science 342, 799.
https://doi.org/10.1126/science.342.6160.799-b.
Hedlund, J., Longo, S.B., York, R., 2020. Agriculture, pesticide use, and economic
The authors declare that they have no known competing financial development: a global examination (1990–2014). Rural. Sociol. 85, 519–544.
interests or personal relationships that could have appeared to influence https://doi.org/10.1111/ruso.12303.
the work reported in this paper. Heimerl, K., Menon, A., Hasan, S., Ali, K., Brewer, E., Parikh, T., 2015. Analysis of
Smartphone Adoption and Usage in a Rural Community Cellular Network, in:
Proceedings of the Seventh International Conference on Information and
Data availability Communication Technologies and Development, ICTD ‘15. Association for
Computing Machinery, New York, NY, USA, pp. 1–4. https://doi.org/10.1145/
2737856.2737880.
The authors do not have permission to share data. Huang, Y., Luo, X., Tang, L., Yu, W., 2020. The power of habit: does production
experience lead to pesticide overuse? Environ. Sci. Pollut. Res. 27, 25287–25296.
https://doi.org/10.1007/s11356-020-08961-4.
Appendix A. Supplementary data
Huang, Y., Luo, X., Liu, D., Du, S., Yan, A., Tang, L., 2021. Pest control ability, technical
guidance, and pesticide overuse: evidence from rice farmers in rural China. Environ.
Supplementary data to this article can be found online at https://doi. Sci. Pollut. Res. 28, 39587–39597. https://doi.org/10.1007/s11356-021-13607-0.
org/10.1016/j.scitotenv.2024.173867. Jallow, M.F.A., Awadh, D.G., Albaho, M.S., Devi, V.Y., Thomas, B.M., 2017. Pesticide
risk behaviors and factors influencing pesticide use among farmers in Kuwait. Sci.
Total Environ. 574, 490–498.
References Janc, K., Czapiewski, K., Wójcik, M., 2019. In the starting blocks for smart agriculture:
the internet as a source of knowledge in transitional agriculture. NJAS - Wagening. J.
Life Sci. 90–91, 100309 https://doi.org/10.1016/j.njas.2019.100309.
Aker, J.C., 2011. Dial “A” for agriculture: a review of information and communication
Jin, S., Bluemling, B., Mol, A.P.J., 2015. Information, trust and pesticide overuse:
technologies for agricultural extension in developing countries. Agric. Econ. 42,
interactions between retailers and cotton farmers in China. NJAS Wagening. J. Life
631–647. https://doi.org/10.1111/j.1574-0862.2011.00545.x.
Sci. 72–73, 23–32. https://doi.org/10.1016/j.njas.2014.10.003.
Al Zadjali, S., Morse, S., Chenoweth, J., Deadman, M., 2014. Factors determining
Kennedy, P.E., 1981. Estimation with correctly interpreted dummy variables in
pesticide use practices by farmers in the Sultanate of Oman. Sci. Total Environ.
Semilogarithmic equations. Am. Econ. Rev. 71, 801.
476–477, 505–512. https://doi.org/10.1016/j.scitotenv.2013.12.040.
Khan, N., Ray, R.L., Zhang, S., Osabuohien, E., Ihtisham, M., 2022. Influence of mobile
Atreya, K., 2007. Pesticide use knowledge and practices: a gender differences in Nepal.
phone and internet technology on income of rural farmers: evidence from Khyber
Environ. Res. 104, 305–311. https://doi.org/10.1016/j.envres.2007.01.001.
Pakhtunkhwa Province. Pakistan. Technol. Soc. 68, 101866.
Awuor, F.M., Otanga, S.A., 2019. Farmer centered large scale e-surveillance and control
Li, W., Sun, L., 2017. Can Tongxiang Guanxi play a role in executive promotion? An
of crop pests in Kenya. J. Agric. Inform. 10 https://doi.org/10.17700/
empirical test based on the heads of provincial state-owned enterprises. J. Financ.
jai.2019.1.1.500.
Econ. 43, 17-28+40 https://doi.org/10.16538/j.cnki.jfe.2017.01.002.
Balli, H.O., Sørensen, B.E., 2013. Interaction effects in econometrics. Empir. Econ. 45,
Li, H., Wang, C., Chang, W.-Y., Liu, H., 2023. Factors affecting Chinese farmers’
583–603. https://doi.org/10.1007/s00181-012-0604-2.
environment-friendly pesticide application behavior: a meta-analysis. J. Clean. Prod.
Baumüller, H., 2018. The little we know: an exploratory literature review on the utility of
409, 137277 https://doi.org/10.1016/j.jclepro.2023.137277.
mobile phone-enabled services for smallholder farmers. J. Int. Dev. 30, 134–154.
Ma, W., Zheng, H., 2022. Heterogeneous impacts of information technology adoption on
https://doi.org/10.1002/jid.3314.
pesticide and fertiliser expenditures: evidence from wheat farmers in China*. Aust. J.
Birke, F.M., Lemma, M., Knierim, A., 2019. Perceptions towards information
Agric. Resour. Econ. 66, 72–92. https://doi.org/10.1111/1467-8489.12446.
communication technologies and their use in agricultural extension: case study from
Migheli, M., 2017. Land ownership and use of pesticides. Evidence from the Mekong
South Wollo, Ethiopia. J. Agric. Educ. Ext. 25, 47–62. https://doi.org/10.1080/
Delta. J. Clean. Prod. 145, 188–198. https://doi.org/10.1016/j.jclepro.2017.01.045.
1389224X.2018.1524773.
Min, S., Liu, M., Huang, J., 2020. Does the application of ICTs facilitate rural economic
Carolan, M., 2022. Digitization as politics: smart farming through the lens of weak and
transformation in China? Empirical evidence from the use of smartphones among
strong data. J. Rural Stud. 91, 208–216. https://doi.org/10.1016/j.
farmers. J. Asian Econ. 70, 101219 https://doi.org/10.1016/j.asieco.2020.101219.
jrurstud.2020.10.040.
Möhring, N., Ingold, K., Kudsk, P., Martin-Laurent, F., Niggli, U., Siegrist, M., Studer, B.,
Carrión-Yaguana, V.D., Alwang, J., Barrera, V.H., 2020. Promoting behavioral change
Walter, A., Finger, R., 2020. Pathways for advancing pesticide policies. Nat. Food 1,
using text messages: a case study of blackberry farmers in Ecuador. J. Agric. Appl.
535–540. https://doi.org/10.1038/s43016-020-00141-4.
Econ. 52, 398–419. https://doi.org/10.1017/aae.2020.7.
Muto, M., Yamano, T., 2009. The impact of Mobile phone coverage expansion on market
Chowdhury, A., Odame, H.H., Thompson, S., Hauser, M., 2015. Enhancing farmers’
participation: panel data evidence from Uganda. World Dev. 37, 1887–1896.
capacity for botanical pesticide innovation through video-mediated learning in
https://doi.org/10.1016/j.worlddev.2009.05.004.
Bangladesh. Int. J. Agric. Sustain. 13, 326–349.
Nakasone, E., Torero, M., Minten, B., 2014. The power of information: the ICT revolution
Clapp, J., Ruder, S.-L., 2020. Precision technologies for agriculture: digital farming,
in agricultural development. Annu Rev Resour Econ 6, 533–550.
gene-edited crops, and the politics of sustainability. Glob. Environ. Polit. 20, 49–69.
Normile, D., 2013. Vietnam turns back a “tsunami of pesticides”. Science 341, 737–738.
https://doi.org/10.1162/glep_a_00566.
https://doi.org/10.1126/science.341.6147.737.
Cole, S.A., Fernando, A.N., 2021. Mobile’izing agricultural advice technology adoption
Nyamba, S.Y., Mlozi, M.R.S., 2012. Factors influencing the use of mobile phones in
diffusion and sustainability. Econ. J. 131, 192–219. https://doi.org/10.1093/ej/
communicating agricultural information: a case of Kilolo District, Iringa. Tanzania 2,
ueaa084.
6.
Enserink, M., Hines, P.J., Vignieri, S.N., Wigginton, N.S., Yeston, J.S., 2013. The
Pan, Y., Ren, Y., Luning, P.A., 2021. Factors influencing Chinese farmers’ proper
Pesticide Paradox. Science 341, 728–729. https://doi.org/10.1126/
pesticide application in agricultural products – A review. Food Control 122, 107788.
science.341.6147.728.
https://doi.org/10.1016/j.foodcont.2020.107788.
Evans, W.N., Oates, W.E., Schwab, R.M., 1992. Measuring peer group effects: a study of
Pan, D., He, M., Kong, F., 2020. Risk attitude, risk perception, and farmers’ pesticide
teenage behavior. J. Polit. Econ. 100, 966–991. https://doi.org/10.1086/261848.
application behavior in China: a moderation and mediation model. J. Clean. Prod.
Fabregas, R., Kremer, M., Schilbach, F., 2019. Realizing the potential of digital
276, 124241 https://doi.org/10.1016/j.jclepro.2020.124241.
development: the case of agricultural advice. Science 366, eaay3038. https://doi.
Popp, J., Pet, K., Nagy, János, 2013. Pesticide productivity and food security. A review.
org/10.1126/science.aay3038.
Agron. Sustain. Dev. 33, 243–255.
Fan, L., Niu, H., Yang, X., Qin, W., Bento, C.P.M., Ritsema, C.J., Geissen, V., 2015.
Rahman, S., 2003. Farm-level pesticide use in Bangladesh: determinants and awareness.
Factors affecting farmers’ behaviour in pesticide use: insights from a field study in
Agric. Ecosyst. Environ. 95, 241–252. https://doi.org/10.1016/S0167-8809(02)
northern China. Sci. Total Environ. 537, 360–368. https://doi.org/10.1016/j.
00089-0.
scitotenv.2015.07.150.
Rahman, S., 2016. Jointness in farmers’ decision to apply pesticides in multiple crops
Fan, P., Mishra, A.K., Feng, S., Su, M., 2023. The effect of agricultural subsidies on
and its determinants at the farm level in Bangladesh. Int. J. Pest Manag. 62,
chemical fertilizer use: evidence from a new policy in China. J. Environ. Manage.
149–157. https://doi.org/10.1080/09670874.2016.1147621.
344, 118423 https://doi.org/10.1016/j.jenvman.2023.118423.
Rotz, S., Duncan, E., Small, M., Botschner, J., Dara, R., Mosby, I., Reed, M., Fraser, E.D.
Folitse, B.Y., Manteaw, S.A., Dzandu, L.P., Obeng-Koranteng, G., Bekoe, S., 2019. The
G., 2019. The politics of digital agricultural technologies: a preliminary review.
determinants of mobile-phone usage among small-scale poultry farmers in Ghana.
Sociol. Rural. 59, 203–229. https://doi.org/10.1111/soru.12233.
Inf. Dev. 35, 564–574. https://doi.org/10.1177/0266666918772005.
Salazar, C., Rand, J., 2020. Pesticide use, production risk and shocks. The case of rice
Freeman, K., Qin, H., 2020. The role of information and interaction processes in the
producers in Vietnam. J. Environ. Manage. 253, 109705 https://doi.org/10.1016/j.
adoption of agriculture inputs in Uganda. Agronomy 10, 202. https://doi.org/
jenvman.2019.109705.
10.3390/agronomy10020202.
Schreinemachers, P., Grovermann, C., Praneetvatakul, S., Heng, P., Nguyen, T.T.L.,
Goodhue, R.E., Klonsky, K., Mohapatra, S., 2010. Can an education program be a
Buntong, B., Le, N.T., Pinn, T., 2020. How much is too much? Quantifying pesticide
substitute for a regulatory program that bans pesticides? Evidence from a panel
overuse in vegetable production in Southeast Asia. J. Clean. Prod. 244, 118738
selection model. Am. J. Agric. Econ. 92, 956–971. https://doi.org/10.1093/ajae/
https://doi.org/10.1016/j.jclepro.2019.118738.
aaq032.
Sharifzadeh, M.S., Abdollahzadeh, G., Damalas, C.A., Rezaei, R., Ahmadyousefi, M.,
Greene, W.H., 2003. Econometric Analysis. Pearson Education India.
2019. Determinants of pesticide safety behavior among Iranian rice farmers. Sci.
Grovermann, C., Schreinemachers, P., Berger, T., 2013. Quantifying pesticide overuse
Total Environ. 651, 2953–2960. https://doi.org/10.1016/j.scitotenv.2018.10.179.
from farmer and societal points of view: an application to Thailand. Crop Prot. 53,
161–168. https://doi.org/10.1016/j.cropro.2013.07.013.

10
L. Xie et al. Science of the Total Environment 943 (2024) 173867

Stock, J.H., Yogo, M., 2002. Testing for Weak Instruments in Linear IV Regression. Zhang, C., Shi, G., Shen, J., Hu, R., 2015b. Productivity effect and overuse of pesticide in
Technical Working Paper Series. https://doi.org/10.3386/t0284. crop production in China. J. Integr. Agric. 14, 1903–1910. https://doi.org/10.1016/
Šūmane, S., Kunda, I., Knickel, K., Strauss, A., Tisenkopfs, T., des Ios Rios, I., Rivera, M., S2095-3119(15)61056-5.
Chebach, T., Ashkenazy, A., 2018. Local and farmers’ knowledge matters! How Zhang, C., Sun, Y., Hu, R., Huang, J., Huang, X., Li, Y., Yin, Y., Chen, Z., 2018.
integrating informal and formal knowledge enhances sustainable and resilient A comparison of the effects of agricultural pesticide uses on peripheral nerve
agriculture. J. Rural. Stud. 59, 232–241. conduction in China | scientific reports. Sci. Rep. 8, 9621.
Tambo, J.A., Mugambi, I., Onyango, D.O., Uzayisenga, B., Romney, D., 2023. Using mass Zhang, Y., Si, W., Chen, L., Shen, G., Bai, B., Zhou, C., 2021. Determination and dietary
media campaigns to change pesticide use behaviour among smallholder farmers in risk assessment of 284 pesticide residues in local fruit cultivars in Shanghai, China.
East Africa. J. Rural Stud. 99, 79–91. https://doi.org/10.1016/j. Sci. Rep. 11, 9681. https://doi.org/10.1038/s41598-021-89204-5.
jrurstud.2023.03.001. Zhao, T., Zhang, Z., Liang, S., 2020. Digital economy, entrepreneurship, and high-quality
Verger, P.J.P., Boobis, A.R., 2013. Reevaluate pesticides for food security and safety. economic development: empirical evidence from urban China. J. Manag. World 36,
Science 341, 717–718. https://doi.org/10.1126/science.1241572. 65–76. https://doi.org/10.19744/j.cnki.11-1235/f.2020.0154.
Wang, C., Gu, H., 2013. The market vs. the government: what forces affect the selection Zheng, H., Ma, W., 2023. Smartphone-based information acquisition and wheat farm
of amount of pesticide used by China’s vegetable grower? J. Manag. World 50-66+ performance: insights from a doubly robust IPWRA estimator. Electron. Commer.
187-188. https://doi.org/10.19744/j.cnki.11-1235/f.2013.11.006. Res. 23, 633–658. https://doi.org/10.1007/s10660-021-09481-0.
Wang, W., Jin, J., He, R., Gong, H., 2017. Gender differences in pesticide use knowledge, Zheng, L., Qian, W., Liu, Q., Guo, X., 2021. The impact of the new round of farmland
risk awareness and practices in Chinese farmers. Sci. Total Environ. 590–591, 22–28. certification on the ecological protection of cultivated land: taking the application of
https://doi.org/10.1016/j.scitotenv.2017.03.053. chemical fertilizers and pesticides as examples. Chin. Rural Econ. 76–93.
Wen, T., 2016. From Agriculture 1.0 to Agriculture 4.0. China Countrys. Discov, Zheng, L., Li, L., Zhao, Z., Qian, W., 2023a. Does land certification increase farmers’ use
pp. 20–26. of organic fertilizer? Evidence from China. J. Land Use Sci. 18, 39–54. https://doi.
Wooldridge, J.M., 2010. Econometric Analysis of Cross Section and Panel Data. MIT org/10.1080/1747423X.2023.2178536.
press. Zheng, L., Su, L., Jin, S., 2023b. Reducing land fragmentation to curb cropland
Wooldridge, J.M., Wadud, M., Lye, J., 2016. Introductory Econometrics: Asia Pacific abandonment: evidence from rural China. Can. J. Agric. Econ. Can. Agroeconomie
Edition with Online Study Tools 12 Months, 1st edition. Cengage Learning Australia, 71, 355–373. https://doi.org/10.1111/cjag.12335.
Victoria Australia. Zhong, W., Chen, Y., Xie, L., 2023. How does internet use promote joint adoption of
Wu, Y., Xi, X., Tang, X., Luo, D., Gu, B., Lam, S.K., Vitousek, P.M., Chen, D., 2018. Policy sustainable agricultural practices? Evidence from rice farmers in China. Int. J. Agric.
distortions, farm size, and the overuse of agricultural chemicals in China. Proc. Natl. Sustain. 21, 2270244 https://doi.org/10.1080/14735903.2023.2270244.
Acad. Sci. 115, 7010–7015. https://doi.org/10.1073/pnas.1806645115. Zhou, J., Jin, S., 2009. Safety of vegetables and the use of pesticides by farmers in China:
Wyche, S., Steinfield, C., 2015. Why don’t farmers use cell phones to access market evidence from Zhejiang province. Food Control 20, 1043–1048. https://doi.org/
prices? Technology affordances and barriers to market information services adoption 10.1016/j.foodcont.2009.01.002.
in rural Kenya. Inf. Technol. Dev. https://doi.org/10.1080/ Zhu, W., Wang, R., 2021. Impact of farm size on intensity of pesticide use: evidence from
02681102.2015.1048184. China. Sci. Total Environ. 753, 141696 https://doi.org/10.1016/j.
Xie, L., Qiu, Z., You, L., Kang, Y., 2020. A macro perspective on the relationship between scitotenv.2020.141696.
farm size and agrochemicals use in China. Sustainability 12, 9299. https://doi.org/ Zossou, E., Van Mele, P., Vodouhe, S.D., Wanvoeke, J., 2009. The power of video to
10.3390/su12219299. trigger innovation: rice processing in central Benin. Int. J. Agric. Sustain. 7,
Zhang, C., Hu, R., Shi, G., Jin, Y., Robson, M.G., Huang, X., 2015a. Overuse or underuse? 119–129.
An observation of pesticide use in China. Sci. Total Environ. 1–6.

11